Update app.py

app.py

@@ -46,7 +46,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 
 # model_id = "mistralai/Mistral-7B-v0.3"
 
-model_id = "mistral-community/Mixtral-8x22B-Instruct-v0.1-4bit"
+model_id = "Qwen/Qwen2-72B-Instruct"
 
 
 tokenizer = AutoTokenizer.from_pretrained(
@@ -59,11 +59,11 @@ accelerator = Accelerator()
 
 model = AutoModelForCausalLM.from_pretrained(model_id, token= token, 
                                                  # torch_dtype= torch.uint8, 
-                                             # torch_dtype=torch.float16,
-                                              # load_in_8bit=True,
+                                             torch_dtype=torch.float16,
+                                              load_in_8bit=True,
                                              #  # torch_dtype=torch.fl,
-                                             # attn_implementation="flash_attention_2",
-                                             # low_cpu_mem_usage=True,
+                                             attn_implementation="flash_attention_2",
+                                             low_cpu_mem_usage=True,
                                              # device_map='cuda',
                                              # device_map=accelerator.device_map,